KNN算法

KNN算法

KNN算法是機器學習算法中的一個很特殊算法

• 角度1: knn算法不用fit操作就能直接預測
• 角度2: knn的數據集其實就直接是fit後的結果, 可以直接預測數據

先手動實現一個knn

import numpy as np
from math import sqrt
from collections import Counter

class KNNClassifier:

    def __init__(self, k):
        '''k 爲knn的個數'''
        self.k = k
        self._X_train = None
        self._y_train = None

    def fit(self, X, y):
        '''X 爲樣本 因爲分類結果'''
        assert X.shape[0] == y.shape[0], '樣本和標籤數量不一致'
        assert self.k <= X.shape[0], '樣本數量級必須大於指定的k'
        self._X_train = X
        self._y_train = y
        return self

    def predict(self, X_predict):
        '''y_predict 爲預測結果'''
        assert self._X_train is not None and self._y_train is not None, 'must fit before predict!'
        assert X_predict.shape[1] == self._X_train.shape[1], '預測數據特徵數必須等於訓練數據特徵數'

        y_predict = [self._predict(x_predict) for x_predict in X_predict]
        return np.array(y_predict)

    def _predict(self, x):
        '''實際預測算法'''
        # 最核心的部分----歐拉距離
        distances = [sqrt(np.sum((x_train - x) ** 2)) for x_train in self._X_train]
        nearSet = np.argsort(distances)
        topK_y = [self._y_train[i] for i in nearSet[:self.k]]
        votes = Counter(topK_y)
        return votes.most_common(1)[0][0]

    def __repr__(self):
        return "KNN(k=%d)" % self.k

用jupyetr拿鳶尾花數據集測試一下效果

from sklearn import datasets
digits = datasets.load_digits()
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(digits.data, digits.target)
# 數據準備完成
%run kNN.py	#映入模塊

#開始預測
my_knn = KNNClassifier(6)
my_knn.fit(X_train, y_train)
y_predict = my_knn.predict(X_test)

print(sum(y_predict == y_test) / len(y_test))

輸出: